Skip to content
#small language model Open access

"Finer Grains Are Stronger" Fails, and Not Because the Theoretical Strength Was Reached ── the prediction at 10 nm, 1550 MPa, is only 14.09 per cent of the theoretical 11000 MPa ── the formula reaches theoretical strength at a grain size of 0.187652 nm, smaller than the 0.256 nm atomic spacing ── what acts is the grain-boundary volume fraction, 30 per cent at 10 nm ── [Paper 397]

Sep 2026 · Zenodo (CERN European Organization for Nuclear Research)
Microstructure and mechanical properties

Abstract

The Hall--Petch relation says finer grains are stronger, and it reverses at 10 to 30 nm. This paper shows that the reversal is not because the theoretical strength was reached──the prediction there does not even reach 14 per cent of it. No new theorem or law is claimed. Scope of this paper (scope note): No new theorem or law is claimed──the Hall--Petch relation, its inverse regime, grain-boundary sliding and estimates of theoretical strength are all known in materials science. No material is designed──alloy composition and processing are not given. Only the numbers inside one formula are examined. No mechanism for the reversal is claimed──which of grain-boundary sliding, grain rotation or dislocation-source starvation dominates is disputed, and this paper does not decide it. Only that it is not the theoretical strength. The constants are not claimed as measurements──sigma_0=50 MPa and k=0.15 MPa. m^1/2 are representative values with copper in mind. No definition of theoretical strength is built──E/10 is a classical estimate; a more careful value depends on crystal and plane. Ductility is not treated──only strength; the loss of ductility with grain refinement is not the subject. Relation to earlier papers: Paper 386 showed that the rate moves with the solubility fixed──here a different quantity begins to act while the strength formula stays put. Paper 389 showed that length, not thickness, is what binds──here it is the volume fraction, not the theoretical strength. Paper 385 showed that the grip term makes the maximum──here the boundary term makes the failure. Paper 300 showed that same-root and different-root can be decided──dislocation strengthening and boundary softening are different-root by that criterion. What is added is putting numbers on strength and volume fraction at six grain sizes, showing that 10 nm reaches only 14.09 per cent of theoretical strength, showing that the formula reaches it only below the atomic spacing, obtaining 39 atoms across a 10 nm grain, and placing the separator at “dislocations or boundary sliding”. First, sigma=sigma_0+k/sqrt(d) diverges as d->0; the formula has no ceiling of its own (Section 1). Second, the prediction at d=10 nm is 1550 MPa (Section 2). Third, and this is the core of the paper. That is only 14.09 per cent of the theoretical strength E/10=11000 MPa (Section 2). Fourth, the formula reaches theoretical strength at 0.187652 nm, smaller than copper’s 0.256 nm atomic spacing (Section 3). Fifth, at 10 nm the grain boundary occupies 30 per cent of the volume; at 100 nm it is 3 per cent (Section 3). Sixth, the separator is whether deformation is carried by dislocations or by grain-boundary sliding (Section 5). the Hall--Petch relation says finer grains are stronger, and it reverses at 10 to 30 nm──but that reversal is not because the theoretical strength was reached. The formula sigma=sigma_0+k/sqrt(d) diverges as d->0 and has no ceiling of its own──so the plausible explanation is that it hit the theoretical strength. This paper checks that with numbers. The prediction at 10 nm, 1550 MPa, is only 14.09 per cent of the theoretical E/10=11000 MPa──it fails with more than a factor of seven left in hand. Even at 20 nm it is 10.10 per cent, so nowhere in the reversal band is theoretical strength close. The boundary volume fraction, meanwhile, reaches 0.3000 at 10 nm against 0.0300 at 100 nm. Strength grows as 1/sqrt(d) and the fraction as 1/d, so the finer the grain the faster the fraction takes over. And the formula reaches theoretical strength only below the size of one atom──d=(k/(E/10-sigma_0))^2=0.187652 nm, smaller than copper’s 0.256 nm atomic spacing, which is to say physically unreachable. “It saturates at the theoretical strength” therefore does not hold: the grain becomes smaller than an atom before the ceiling arrives. The observed reversal at 10 to 30 nm is between 53 and 160 times that size, so the plausible explanation was off by orders of magnitude. As in Paper 389, the quantity that gets named is not the one that acts. What fails is not the formula but its premise──Hall--Petch presumes dislocation pile-up, and a 10 nm grain, thirty-nine atoms across, holds only a few dislocations, so the picture of a pile-up stops making sense. The boundary itself starts carrying the deformation when three tenths of the volume is boundary. The formula remains correct; one has simply left the region where it applies. By the criterion of Paper 300 they are different-root: dislocation strengthening inside the crystal, boundary softening at the interface. And the volume fraction does not appear in the formula at all──a quantity absent from the formula beginning to act is what the failure is. One thing does the separating──does deformation happen inside the crystal or at the interfaces. Above 100 nm dislocations lead and refining strengthens; below 10 nm boundary sliding leads and refining weakens. The boundary between the two regions is set by the volume fraction, three per cent against thirty, not by any absolute value of the strength. In a material with a thinner boundary the reversal should occur at a finer grain size, which is a testable prediction. To be honest──both sigma_0=50 MPa and k=0.15 are representative values with copper in mind and differ by material. What does not move is that the grain size at theoretical strength is smaller than the atomic spacing, and that the volume fraction grows as 1/d. One last thing──“it stopped because it hit the ceiling” is most suspect precisely when the ceiling is in view. Put in the numbers and there was still a factor of seven to go. What stopped it was a quantity that is not in the formula. On the making of this work: The ideas and content of this work stem from the author's own considerations. Assistance from an AI (a large language model) was used for structuring, English translation, and checking the algebra. Any remaining errors or misinterpretations are solely the author's. Feedback and corrections are sincerely appreciated. ----- ホール・ペッチ則は粒径が小さいほど強度が上がると言い、10〜30 nm で逆転する。本稿が示すのは逆転が理論強度に達したせいではないことである──そこでの予測値は理論強度の 14 パーセントにも届いていない。新しい定理も法則も主張しない。 本稿の射程(射程注記):新しい定理も法則も主張しない──ホール・ペッチ則、その逆転、粒界すべり、理論強度の見積もりはすべて材料科学で既知である。材料を設計しない──合金組成や加工法は与えない。一つの式の中の数だけを見る。逆転の機構を主張しない──粒界すべり・粒界回転・転位源の枯渇のどれが主かは論争があり、本稿は決めない。「理論強度ではない」ことだけを示す。定数を実測と主張しない──sigma_0=50 MPa、k=0.15 MPa. m^1/2 は銅を念頭に置いた代表値である。理論強度の定義を作らない──E/10 は古典的な見積もりであり、より精密な値は結晶と面に依る。延性を扱わない──強度だけを見る。細粒化に伴う延性の低下は主題ではない。既刊との関係:論文386 は溶解度を動かさずに速さが動くことを示した──ここでは強度の式が動かないまま、別の量が効き始める。論文389 は厚さではなく長さが効くことを示した──ここでは理論強度ではなく体積分率が効く。論文385 は掴みの項が最大を作ることを示した──ここでは粒界の項が破れを作る。論文300 は同根か別根かが判定できることを示した──転位の強化と粒界の軟化は、その基準で別根である。加えたのは、六つの粒径で強度と体積分率を数に出したこと、10 nm で理論強度の 14.09 パーセントにとどまることを示したこと、式が理論強度に達する粒径が原子間距離より小さいことを示したこと、10 nm の粒が原子 39 個ぶんであることを出したこと、分離子を「転位か粒界すべりか」に置いたことである。 第一に、sigma=sigma_0+k/sqrt(d) は d->0 で発散する。式の側に上限が無い(第1節)。 第二に、d=10 nm での予測値は 1550 MPa である(第2節)。 第三に、これが本稿の芯である。それは理論強度 E/10=11000 MPa の 14.09 パーセントでしかない(第2節)。 第四に、式が理論強度に達する粒径は 0.187652 nm で、銅の原子間距離 0.256 nm より小さい(第3節)。 第五に、10 nm では粒界が体積の 30 パーセントを占める。100 nm では 3 パーセントである(第3節)。 第六に、分離子は「変形が転位で担われるか、粒界すべりで担われるか」である(第5節)。 ホール・ペッチ則は粒径が小さいほど強度が上がると言い、10〜30 nm で逆転する──だがその逆転は、理論強度に達したせいではない。式 sigma=sigma_0+k/sqrt(d) は d->0 で発散し、式の側に上限が無い──もっともらしい説明は「理論強度に頭を打ったから」である。本稿はそれを数で確かめる。10 nm での予測値 1550 MPa は、理論強度 E/10=11000 MPa の 14.09 パーセントでしかない──七倍以上の余地を残したまま破れている。逆転域の 20 nm でも 10.10 パーセントで、どこを取っても理論強度には遠い。一方、粒界の体積分率は 10 nm で 0.3000 に達し、100 nm の 0.0300 とは一桁違う。強度は 1/sqrt(d)、体積分率は 1/d で育つので、細かくするほど後者のほうが速く効く。そして式は、原子より小さい粒でしか理論強度に達しない──d=(k/(E/10-sigma_0))^2=0.187652 nm で、銅の原子間距離 0.256 nm より小さい。つまり物理的に到達できず、「理論強度で頭打ちになる」という説明は成立しない──頭打ちの前に、粒が原子より小さくなる。実際の逆転は 10〜30 nm で、0.19 nm とは約 100 倍離れており、もっともらしい説明が桁で外れていた。論文389 で厚さではなく長さが効いていたのと同じ形で、名前の挙がっている量が効いていない。破れているのは式ではなく、式の前提である──ホール・ペッチ則は転位の堆積を前提にしており、10 nm の粒(一辺が原子 39 個ぶん)には転位が数本しか入らず、堆積という描像が成り立たなくなる。代わりに体積の三割を占める粒界そのものが変形を担い始める。式は正しいまま、適用できる領域から出ただけである。論文300 の基準で別根で、転位の強化は結晶内部が、粒界の軟化は界面が決める。そしてこの式には体積分率がそもそも入っておらず、入っていない量が効き始めることが破れの正体である。分けているものは一つ──変形の主役が結晶の内部にあるか、界面にあるか。100 nm より粗ければ転位が主役で細かくするほど強くなり、10 nm より細かければ粒界すべりが主役で細かくするほど弱くなる。境目を決めているのは体積分率(3 パーセントと 30 パーセント)であって、強度の絶対値ではない。粒界の厚みが小さい材料なら逆転はもっと細かい粒径で起きるはずで、これは検証できる予測である。正直に言えば──sigma_0=50 MPa も k=0.15 も銅を念頭に置いた代表値であり、材料ごとに変わる。動かないのは、式が理論強度に達する粒径が原子間距離より小さいことと、体積分率が 1/d で育つことである。最後に一つ──「上限に頭を打ったから止まった」という説明は、上限が見えているときほど疑わしい。数を入れれば、上限までまだ七倍あった。止めたのは、式に入っていない量である。 作成にあたって:本稿の着想と内容は、著者自身の考察に基づくものです。文章の構成整理や英訳、数式の確認には AI(大規模言語モデル)の助力を得ました。最終的な内容の解釈や誤りがあれば、それらはすべて著者の責に帰します。お気づきの点があれば、ご教示いただければ幸いです。

View source

Similar papers

#small language model Dataset Open access Oct 2026

Socratic guiding questions in synthetic arithmetic data: matched LoRA runs (revision v2)

Supporting data, adapters, predictions and code for the article *Low-Cost LoRA Fine-Tuning of Small Language Models for Multi-Step Arithmetic Reasoning* by Jake O'Grady, Asena Isik Gürhan, Chee Fong Ting and Effirul Ramlan (University of Galway). We generated 20,000 GSM8K-derived arithmetic problems with step-by-step s...

O'Grady, Jake, Gürhan, Asena Isik, Chee, Fong Ting et al. · 465 citations
#computer vision Open access Jun 2016

Software Development in Startup Companies: The Greenfield Startup Model

The results are packaged in the Greenfield Startup Model (GSM), which explains the priority of startups to release the product as quickly as possible, and the need to shorten time-to-market, by speeding up the development through low-precision engineering activities.

Carmine Giardino, Nicolò Paternoster, M. Unterkalmsteiner et al. · 178 citations · ⚡14
#computer vision Open access Oct 2016

Software Startups - A Research Agenda

Software startup companies develop innovative, software-intensive products within limited timeframes and with few resources, searching for sustainable and scalable business models.

M. Unterkalmsteiner, P. Abrahamsson, Xiaofeng Wang et al. · 157 citations · ⚡17
#machine learning Review Open access Oct 2016

“Failures” to be celebrated: an analysis of major pivots of software startups

This study conducts a case survey study based on the secondary data of the major pivots happened in 49 software startups, and demonstrates that customer need pivot is the most common among all pivot types.

Sohaib Shahid Bajwa, Xiaofeng Wang, Anh Nguyen-Duc et al. · 127 citations · ⚡15
#computer vision Review Open access May 2015

A survey study on major technical barriers affecting the decision to adopt cloud services

The comparison of adopter and non-adopter sample reveals three potential adoption inhibitor, security, data privacy, and portability, which underlines the importance of the technical and security perspectives for research investigating the adoption of technology.

Nattakarn Phaphoom, Xiaofeng Wang, S. Samuel et al. · 111 citations · ⚡8
#computer vision Open access Feb 2018

Lean Internal Startups for Software Product Innovation in Large Companies: Enablers and Inhibitors

This study investigates how Lean internal startup facilitates software product innovation in large companies and identifies its enablers and inhibitors, and shows the potential of the method-in-action framework to investigate the Lean startup approach in non-startup context.

Henry Edison, Nina M. Smørsgård, Xiaofeng Wang et al. · 78 citations · ⚡6

Related blog posts

MIT News · Artificial Intelligence Oct 15, 2026

AI model decodes the language cells use to communicate

A new computational approach reveals how cells communicate to assemble into tissues. This could accelerate stem cell engineering, improve disease models, and inform therapies.

Microsoft Research Blog Oct 15, 2026

Proof-based general-purpose computing with Magnetar

Zero-knowledge proofs can verify computing results without rerunning workloads or exposing sensitive data. Magnetar helps bring them into the mainstream with support for standard programming languages, major speed gains, and quantum-resistant security. The post Proof-based general-purpose computing with Magnetar appeared first on Microsoft Research.

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.